Bought a faulty Sega Rally twin which had graphics glitches on one side. The fault manifested itself with random polygons disappearing from view, and when switching between viewpoints sometimes a few of the polygons from the external car camera would be left in the 1st person camera.
First I noted that the boardset in the working side had had a previous repair on the video board, a 256kBit static ram at IC2 (TC55328AP-20), had been replaced, so I duplicated this repair on the faulty board, but alas it had no effect on the fault.
The boards were heavily endowed with dust from the cage fan, which I know from Daytona USA repairs usually indicates one or more of the surface mount memories has gone, or some of the main customs need reflowing. I decided against reflowing all the chips in the area and tried replicating the fault by taking the known working boardset I had, and grounding the databus pins on the memory chips spread around the video board. While I don't recommend this as you can easily hit a 5v line or might cause damage, I have a history of poking probes where I shouldn't so I went on a rampage.
I found that grounding certain databus pins on the 64Kbit static ram chip at IC4 (CY7C185 25ns) caused similar loss of polygons, so I took the same chip off a known dead model 2A video board and fitted it to the faulty board, and hey presto, fixed!
Here's a video of the fault in action, apologies for the low bitrate.
IC4 near the middle here:
And the end result (the bench monitor needs recapping):
First I noted that the boardset in the working side had had a previous repair on the video board, a 256kBit static ram at IC2 (TC55328AP-20), had been replaced, so I duplicated this repair on the faulty board, but alas it had no effect on the fault.
The boards were heavily endowed with dust from the cage fan, which I know from Daytona USA repairs usually indicates one or more of the surface mount memories has gone, or some of the main customs need reflowing. I decided against reflowing all the chips in the area and tried replicating the fault by taking the known working boardset I had, and grounding the databus pins on the memory chips spread around the video board. While I don't recommend this as you can easily hit a 5v line or might cause damage, I have a history of poking probes where I shouldn't so I went on a rampage.
I found that grounding certain databus pins on the 64Kbit static ram chip at IC4 (CY7C185 25ns) caused similar loss of polygons, so I took the same chip off a known dead model 2A video board and fitted it to the faulty board, and hey presto, fixed!
Here's a video of the fault in action, apologies for the low bitrate.
IC4 near the middle here:
And the end result (the bench monitor needs recapping):
